DDAPRED

DDAPRED predicts novel therapeutic indications for existing drugs to support drug repositioning by integrating drug and disease similarity information with regularized logistic matrix factorization.

Key Features:

• Regularized Logistic Matrix Factorization: Employs regularized logistic matrix factorization (also described as regularized logistic matrix decomposition) to model drug-disease associations.
• Integration of Multiple Data Sources: Integrates drug similarity and disease similarity datasets to capture complex relationships between drugs and diseases.
• Performance Metrics: Reported 5-fold cross-validation performance with AUROC of 0.932 and AUPRC of 0.438.
• Parameter Analysis: Includes analysis of model parameters affecting predictive performance.
• Validation of Predictions: Validated top 50 predicted drug-disease pairs by analyzing their treatment relationships for previously unknown associations.

Scientific Applications:

• Drug Repositioning Prediction: Predicts potential new indications for existing pharmaceuticals to prioritize candidates for further validation.
• Precision Medicine: Identifies novel drug-disease associations that can inform tailored therapeutic strategies for specific disease profiles.

Methodology:

Applies regularized logistic matrix factorization (referred to as regularized logistic matrix decomposition) to integrate drug similarity and disease similarity matrices and predict drug-disease associations, with evaluation by 5-fold cross-validation.